Preventing and Minimizing...

75

C H A P T E R T W O

“Any sudden event which creates a great demand for actual cash maycause, and will tend to cause, a panic in a country where cash is mucheconomized, and where debts payable on demand are large.”

Walter Bagehot (1873)

WHEN MAJOR FINANCIAL CRISES OCCUR, ALL

who depend on financial services suffer.Depositors can lose their funds or havetheir accounts frozen and value eroded byinflation. Good borrowers get cut off fromcredit. Issuers of debt and equity finance find

that markets have dried up. Pensioners may find their living standardsdiminished. Holders of insurance policies may find their counterpartybankrupt. And taxpayers often foot a bill that otherwise could havepermitted much-needed expenditures on other items. Even those so poorthat they do not use the financial services of the formal sector may findtheir incomes slashed in the resulting recession, and informal financialfunds may dry up as well (box 2.1).

Recent decades have seen a record wave of crises: by millennium-end,there had been 112 episodes of systemic banking crises in 93 countriessince the late 1970s—and 51 borderline crises were recorded in 46 coun-tries. These crises both were more numerous and expensive, comparedwith those earlier in history, and their costs often devastating in develop-ing countries.

This chapter first examines why finance is so fragile—especially indeveloping countries, and all the more so in banking—and it discussesthe costs of financial and banking crises, and their causes. Banking crises

Preventing and MinimizingCrises

Recent financial criseshave been more numerousand expensive than in thepast

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0475

F I N A N C E F O R G ROW T H : P O L I C Y C H O I C E S I N A V O L AT I L E W O R L D

76

are the main focus, and although currency misalignment is a commonelement of a banking crisis, so-called twin (banking cum currency) cri-ses are deferred until chapter 4.

How can society be provided with financial services without incurringthe costs of these crashes? The incentive structure, the product of marketforces interacting with the regulatory environment, is undoubtedly thekey factor in the stability and functioning of the financial sector, so thesecond section of the chapter, Regulating Banks: Harnessing the Market,turns to reform in this area. Just as liberalization of private initiative in thefinancial sector and real and financial technological developments havebeen part of the story in the increased vulnerability of finance in recentdecades, so creative initiatives to harness the private sector and technologyare key to bringing the social risks of finance back under control.

WHEN CRISES OCCUR AND LENDERS BECOME MORE

risk averse, small firms are the first to be rationedfrom access to credit, which is an important reasonwhy small business failure rates soar during financialcrises. Not surprisingly, then, poverty can rise sharplyand remain high for some time following a crisis.

Number of people living in poverty

RepublicYear Indonesia of Korea Thailand

1990 80.9 14.7 18.4a

1996 50.6 4.7 7.51998 9.1 7.61999 76.3 9.72000 70.3 6.0 8.7

Note: Figures for 2000 are estimates.a. 1988 data.Source: ???

Even with the recovery and projected declinein poverty rates in 2001, the number of poor

people is expected to return to precrisis levels onlyin Thailand, and remain high in Indonesia andthe Republic of Korea. As serious as this impactis, the poor get hit again when the bill comes due,as loan losses sooner or later have to be covered(figure 2.1). Fiscal costs of bank insolvency, whichrepresent injections of government funds, must becovered by tax increases, expenditure reductions,or inflation, all of which hit low-income house-holds hard. Even if authorities attempt to put oncontrols to prevent capital flight, experience showsthat wealthy households are best able to avoidthem; middle- and low-income families’ funds arethen left to bear the burden of higher taxes, soincome distributions usually deteriorate for at leastseveral years after a crisis. Subsequent growth“…tends not to eliminate the higher level of in-equality generated during a severe economic down-turn” (Lustig 1999). Consequently, preventing fi-nancial crises is an important and potentiallyeffective instrument to sustain growth and avoidpoverty.

Box 2.1 Poverty and crises

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0476

77

P R E V E N T I N G A N D M I N I M I Z I N G C R I S E S

A key facet of the incentive environment is the safety net provided forbanks. The 20th century was marked by the rise of safety nets for thebanking sector, the main components of which are the lender of last resortfacility and deposit insurance. Although much has been written on theformer, research on deposit insurance has been mostly theoretical and lim-ited to the United States until recently. Given the recent expansion ofexplicit deposit insurance systems around the world, we then focus in thethird section, Financial Sector Safety Nets, on when and how they canbest be designed. An excessively generous safety net for banks—or stateownership, discussed in chapter 3—can be a key factor behind the bankdominance and the fragility in many emerging markets.

Finance is anything but static: once a set of rules is promulgated, thenature of finance makes it especially easy for participants to move theirbusiness into different forms or jurisdictions that can nullify the goals ofreforms. This regulatory arbitrage will vary directly with the extent towhich regulations neglect the optimizing behavior of participants. Fi-nancial systems in which incentives encourage prudent risk-taking will,other things equal, be more resilient, less a source of shocks, and there-fore better able to assist in risk mitigation. And as incentive-compatibleregulation is combined with an infrastructure that encourages efficientmarket functioning, economic growth will be stimulated by intermedi-aries with the incentives and wherewithal to engage in prudent risk-taking. This does not mean relying naively on markets to do the job, butrather shaping incentives of private agents and regularly revisiting theeffects of various changes on them, what might be termed dynamic regu-lation. To understand better the consequences of the current regulatoryenvironment, as well as the advantages and disadvantages of any reforms,authorities must focus on the underlying incentives.

It may be necessary to go further than setting out a program of regu-latory reform and safety net issues in this area. We have to ask whetherthere are deeper reasons why such reform has not long since been put inplace in most countries. Is it really a failure of regulatory design, or couldit also reflect weakness in the political institutions? Is it in the interest ofsome interest groups and their political sponsors that a lax regulatoryenvironment and a safety net with perverse incentive effects be main-tained even though they increase the risk of socially costly bank failure?That issue goes beyond the scope of this chapter, and indeed beyondmuch research, though we return to related matters in chapter 3.

The incentive structure iskey to the stability andfunctioning of the financialsystem

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0477


78

Why Finance Has Been So Fragile...and RemainsThat Way

All people are most credulous when they are most happy; and whenmuch money has just been made, when some people are really makingit, when most people think they are making it, there is a happy oppor-tunity for ingenious mendacity. Almost everything will be believed fora little while, and long before discovery the worst and most adroitdeceivers are geographically or legally beyond the reach of punishment.But the harm they have done diffuses harm, for it weakens credit stillfurther.

Walter Bagehot (1873, p. 151)

IN PERFORMING ITS ESSENTIAL FUNCTIONS, FINANCE REGULARLY

involves the exchange of money today for the promise of moneyin the future, usually with some form of return. This intertemporal

nature, combined with well-known information problems that admitadverse selection and moral hazard behavior, is at the heart of the fragilityof finance. Each party to this trade enters into the contract with expectationsabout a host of variables that will affect the likelihood of repayment.Expectations change, perhaps quickly, and lead to swings in asset prices,which in turn may be exacerbated by the possibility of crowd behavior.

To be sure, there is some truth in the idea that financial marketsnormally make a reasonably efficient use of information in the sensethat it is hard for an investor consistently to earn excess returns—atleast on a risk-adjusted basis—using publicly available information.Indeed, even information that is not widely available can quickly be-come embodied in market prices as long as there are enough well-financed, informed investors.

Although the “efficient markets” hypothesis is a useful benchmarkfor describing the evolution of market prices in normal times, it ishard-pressed to explain the scale of price movements in turbulent con-ditions. Although itself more than a fad, stock in the efficient marketshypothesis “...crashed along with the rest of the market on October19, 1987. Its recovery has been less dramatic than that of the rest ofthe market” (Shleifer and Summers 1990, p. 19). Indeed, there aresound theoretical reasons why financial markets cannot be efficientand fully arbitraged if information is less than perfect and contractingis costly (Grossman and Stiglitz 1980). Substantial and even growing

The efficient markethypothesis can not explain

speculative booms andbusts

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0478

79


deviations from equilibrium prices are possible, manifesting themselvesas bubbles, or speculative booms and busts. And bubbles are morelikely when, as is found in experiments, individuals are not fully ratio-nal in assessing risk; excessively weight recent experience (display myo-pia); trade on noise rather than on fundamentals; or exhibit positivefeedback (or momentum) by buying because prices are rising.1

The “behavioral finance” view that asset markets are prone to bubblesfinds confirming evidence in countless episodes of sudden asset pricecollapses, with greater or less involvement by the banking sector. Anaugmented and updated version of Kindleberger’s (1978) list (table 2.1)shows the regularity of major incidents since the 15th century, as wellas the diversity of the objects of speculation. Real estate, a commonstumbling block for banks in the latter half of the 20th century, hasearlier antecedents in the list, but there are also many other targets fromcommodities—mineral, such as copper, silver, and gold, or even veg-etable; to mines; all sorts of company shares, financial and nonfinancial,notably utilities such as canals and railroads; and latterly paper moneyand financial derivatives.

Ponzi, or pyramid, schemes, in which investors are gulled into givingfunds to nefarious characters who promise impossibly high rates of re-turn (typically rationalized through complex, apparently “fail safe” means)also illustrate the characteristic fragility of finance.2 These schemes gaincredibility by actually paying the promised returns to early investors outof the cash generated from later investors. Although it is doubtful thatthere is a country that has not seen these schemes, their occurrence in somany transition economies in the 1990s testifies to their link to opaqueenvironments and times of structural change. In some cases, such as theRomanian pyramid of the mid-1990s, railroad traffic even in other coun-tries was said to be affected by the rush to get to the town of Cluj, whereinvestors could get into a scheme promising to repay 8-fold in 100 days—an annual rate of return of 250,000 percent. The scheme collapsed shortlybefore threatening to overtake Romanian GDP, notwithstanding the fact,relatively unique for these schemes, that there was not even a clear storyof how the funds were to be invested.3 Shortly thereafter, Albania saw aseries of schemes the aggregate size of whose liabilities rose to an esti-mated 50 percent of GDP and whose collapse led to widespread streetviolence and 2,000 casualties.

If finance is fragile, banking is its most fragile part, for it adds thecomplications, not only of maturity transformation, but of demandable

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0479


80

Table 2.1 Selected financial crashes (grouped by the object of speculation)

Commodities Companies Real Estate Banks Financial Assets

1400 Bardi & Peruzzi(Florence), 1348

1500 Gold (New World), Medici (Florence), Bourse loans1550s 1492 (Antwerp), 1557

1600 Coins in Spain, 1618 Dutch East India Canals, elegant Fugger (Augsburg),Co., 1636–40 houses (Holland) 1596

1636–40Tulips, 1640

1700 South Seas Sword Blade(London), Companie (London), Banques

d’Occident Generale & Royale(Paris), 1720 (Paris), 1720

British country British gilts inbanks, 1750s Amsterdam 1763

British and DutchEast India Co.,

1772Dutch East India

Co., 1783Sugar, coffee, 1799 French canals, 1793 British country Assignats

banks, 1793 (France), 17951800 Exports, 1810 and Biens Nationaux

1816 (France), 1825British, French Chicago, 1830–42 British country Foreign bonds,canals, 1820s banks, 1824 foreign mines,

new companies,Britain, 1825

Cotton in Britain, British railroads, Chicago, 1843–62France; exports in 1836

Britain 1836Sugar, coffee in British and French Chicago, U.S. Germany, 1850 Foreign mines,

Hamburg, wheat, railroads, 1847 public land, Britain, France 18501857 1853–77

Cotton, 1861 French and U.S. Overend Gurneyrailroads, 1857 (London), 1866;

Gold (New York), Credit Mobilier1869 (Paris), 1867

Petroleum (U.S.), U.S. railroads, Chicago, Berlin, Germany 1870s1871 1873 Vienna, 1878–98

Copper (France); Panama Canal Argentine public Union Generale Foreign bonds,1888; Petroleum Company, France, lands; Chicago, (Paris), 1882 France; British(Russia), 1890s 1888 1890s discount houses, 1888

U.S. railroads, Barings (London),1893 1890

(table continues on following page)

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0480

81


Table 2.1 (continued)

Commodities Companies Real Estate Banks Financial Assets

1900 Copper, U.S., 1907 KnickerbockerTrust (New York),

1907International Bills of Exchange,Mercantile London, 1914

Marine, 1914General Motors, U.S. farmland, Creditanstalt 1920s: German

1920 1918–21 (Austria), 1931 reichsmark,French franc

Florida, 1920s 500 U.S. banks, Mergers, U.K.;1932–33 foreign bonds, new

shares, N.Y.Penn Central FDI, U.S.

Railroad, 1970 conglomerates,sterling, 1960s

Oil tankers, 1974 Burmah Oil, 1974; U.S. farmland U.S. dollar, 1973Pertamina 1970s

(Indonesia), 1975Gold, 1978–82 Chrysler Auto, U.S. Southwest, Banco Ambrosiano LDC debt

1979 California (Italy), 19821970s–80s

Silver, 1980 U.S. S &Ls, 1980s U.S. dollar (1985)Argentina, FDI in U.S., 1980s1980–89

Chile 1981 Junk bonds (U.S.),1989–90

Coffee, cocoa etc., U.S. REITs, offices, Japan, U.S. Japanese shares;1986 malls, hotels; Japan, 1980s–92 Vietnamese credit

Sweden 1980s cooperativesSweden 1990 1990s: Korean

mergersPanAmerican BCCI, 1991 Emerging marketAirways, 1991 shares, 1990sGuinness Peat Romanian, AlbanianAviation, 1992 Ponzi Schemes

Mexico 1994Copper, Japan 1996 Barings Derivatives (Orange

(Singapore), 1995 County;Metallgesellschaft,

Ashanti Gold Mines),forex futures, options

Korean Chaibols; Thailand, Indonesia, Republic Russian bonds, longThailand 1997 1996–97 of Korea, Malaysia, term capital

Thailand 1997–98 management, 1998High tech stocks,

U.S. dollar 1997–??

Note: Items in italics indicate government support and items in bold indicate a major crash.Source: ???

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0481


82

debt, that is, offering debt finance backed by par value liabilities in theform of bank deposits. This particularly fragile structure of its liabilitiesmay be needed to keep the bankers on their toes and to give large de-positors the comfort that they can withdraw as soon as they suspectproblems. Banks arose precisely to finance relatively illiquid investmentswith mostly short-term liabilities (and the fragility of their liability struc-ture has been seen by some scholars as an essential part of their make-up—without which paradoxically they might not be able to function atall. Cf. Diamond and Rajan 2000; Calomiris and Kahn 1991).4 It also,however, makes banks—and even the whole banking system—suscep-tible to a sudden withdrawal of deposits. Although all outsiders will havedifficulty in monitoring banks, depositors—other than the largest—arelikely to be weak at monitoring and also will have an incentive to “freeride” on the monitoring efforts of others. Even if insolvent banks are thefirst to see a withdrawal of deposits, the contraction of lending by somebanks can produce legitimate solvency concerns about others to the ex-tent that aggregate credit shrinks. Indeed, even when banks seem to be-have prudently, the bursting of asset bubbles can impair the ability ofdebtors to repay and induce doubts about banks’ health.

Thus, banking may be characterized by the possibility of contagiousruns, in which a run on one bank leads to runs on other, possibly healthy,banks. In contrast, equity mutual funds, which invest in stocks and paya return that varies with the return on their portfolio, may suffer fromsharp swings in prices, but not from the possibility of contagious runs.However, contagious runs, in the sense that healthy banks are broughtdown by failures at weak banks, in fact are difficult to find, at least inindustrial economies. Even during the U.S. Depression, Calomiris andMason (2000) find that individual fundamentals explain the runs of1930 and 1931, but not the 1933 episode, which they link to a general-ized run from dollars because of the expectation of a devaluation. Thefear of contagious runs may be more marked in emerging markets, be-cause of greater information problems, but emerging markets also mayface a greater tendency toward generalized runs, since shocks sufficientlylarge to change macropolicies or affect the solvency of the banking sys-tem are more common (below). And as noted below, the cost of crisisalso involves the ensuing credit crunch, all the more so in economieswithout alternative channels of finance.

The particular fragility of finance, and within it of banking, is truefor all countries regardless of their income level, as attested to by the

Banking is the most fragilepart of a financial

system—

—and a limited crisis mayaffect the whole banking

network through contagion

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0482

83


occurrence of banking crises in several industrial economies in the 1980sand 1990s. Banking outside the industrial world, however, is more dan-gerous still, where crises have been enormously costly (figure 2.1).

The cumulative losses of the failed banks are only one aspect of thecost of a banking crisis. In attempting to arrive at an estimate of thetotal true economic cost it is necessary to distinguish between threekey components:

• The stock component is the accumulated waste of economic re-sources that is revealed by the insolvency. At least part of the capi-tal deficiency of the failed banks represents depositors’ funds thathave been wasted in unrecoverable loans that were applied to un-productive purposes, such as empty offices and closed factories.

Figure 2.1 Total fiscal costs (increases in the stock of public debt)relative to the flow of GDP in the year of crisis

Source: Honohan and Klingebiel (2000); Caprio and Klingebiel (1999).

Sweden

Colombia

Paraguay

Norway

Hungary

Czech Republic

Ecuador

Philippines

Malaysia

Japan

Côte d’Ivoire

Thailand

ChileIndonesia

Republic of Korea

Uruguay

Republica Bolivariana de Venezuela

Mexico

Slovak Republic

Brazil

Bulgaria

Finland

Senegal

Spain

Sri Lanka

Malaysia

0 20 30 504010

Percent of GDP

199119911982198919951995

198719881991199119891996199619941983

19921997

19941992

19941988

19811997

19971981

1997

Banking crises have been costly.

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0483


84

• The public finance component of the true economic costs arisesbecause of the way in which the fiscal authorities tend to assume alarge part of the net capital deficiency of the banks, in order to bailout the depositors and others directly affected by the crash. Froman economic cost calculation, this cash “fiscal cost” is merely atransfer to depositors, but it also entails a deadweight economiccost that could represent a sizable fraction of the amount trans-ferred where the marginal cost of social funds is high. The point isthat the expenditure cuts, additional tax revenue that will be re-quired to finance them, and/or the inflation tax have distortingeffects in themselves, especially in developing countries with weakrevenue-raising systems. Thus, for example, “merely” servicing thedebt incurred as a result of the Indonesian banking crisis meansspending sums that could have doubled health and education spend-ing. Moreover, in many emerging markets, the fiscal costs are suf-ficiently large to derail macroeconomic stabilization programs, withcostly consequences.

• The flow component of the economic cost arises from the outputslumps with which banking crises are almost always associated.This clearly represents an economic cost inasmuch as resources areunderemployed until the economy picks up again. Channelsthrough which this disruption can occur include a collapse of in-vestment and other spending either because of a general loss ofconfidence, or through a restriction of access to credit (reflectingwould-be borrowers being strapped for collateral; lenders’ reactionto the crisis by raising creditworthiness standards or attempts toremain liquid; or the loss of information capital, essential for mak-ing loans).5 Payments system failure, though rare, can be anotherchannel for triggering recession. As well as a transitory dip in out-put below full employment levels, these channels can result in fur-ther loss of trend output if the lack of intermediated credit de-presses long-term productivity growth.

The larger the initial capital deficiency of the failed banks, the largerthe cash fiscal cost and the larger each of the components of the trueeconomic cost is likely to be. Estimates, of varying reliability, of the cashfiscal cost have been made for many crises. Total fiscal costs in developingcountry crises during the 1980s and 1990s breached the $1 trillion dollarlevel by 1999. These fiscal costs likely overstate the fiscal component of

Banking crises have realcosts

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0484

85


true economic costs, but may be used as a general indication of the relativeand absolute magnitudes of total economic costs.

Alternatively, attempts have also been made to capture a rough esti-mate of the additional flow economic costs, typically by comparing ac-tual output with some hypothetical “no crisis” output path. It is veryhard, though, to guess what part of an output slump is caused by thebanking crisis—often a latent banking crash only becomes evident whenit is triggered by an exogenous economic shock that also directly con-tributed to recession. The measured output dip likely overstates the trueflow economic costs, but it is correlated with measured fiscal costs, andintriguingly is of the same order of magnitude (figure 2.2).6 As Boydand Smith (2000) observe, many crises, though serious at the time, havea small fiscal cost and a relatively low output cost. In figure 2.2, how-ever, about one crisis in three has a cumulative GDP cost of 20 percentor more, and given the uncertainty in times of crises, authorities cannotknow whether they will have a small or a large crisis. Given the depth ofthe recessions, the proverb that an ounce of prevention is worth a poundof cure seems applicable.

Developing countries suffer several additional sources of fragility. First,information problems in general are more pronounced, as noted in the

Fiscal and output costsgenerally go hand in hand.

Figure 2.2 Estimates of fiscal cost of and output dip for 39 banking crises

Note: The chart shows that the fiscal cost of crises is correlated with the subsequent output dip(measured as the total output loss—relative to trend—over the period during which growth re-mained below precrisis rates).

Source: Honohan and Klingebiel (2000).

0 10 20 70

Fiscal cost (percent of GDP)

Output dip: cumulative percent of GDP

30 40 50 60

70

60

50

40

30

20

10

0

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0485


86

discussion in chapter 1 on the accounting and legal systems. This infor-mation problem has to be addressed in any recommendations on lessen-ing vulnerability. Poor information makes it easier for banks not just totake risks unwisely, but also to engage in deliberate related lending, whichaccording to both anecdotal evidence and now empirical research (LaPorta, López-de-Silanes, and Zamarripa 2000) is characterized by muchhigher nonrepayment rates.

Second, developing economies are smaller and more concentrated incertain economic sectors or reliant on particular export products, andaccordingly, they are less able to absorb or pool isolated shocks. This inpart explains the greater macroeconomic volatility displayed by devel-oping economies in different parts of the world in comparison with theindustrial countries (figure 2.3).

Since the portfolios of most financial intermediaries in emergingmarkets are overwhelmingly concentrated in domestic assets, shocks tothe local economy would be more destabilizing even with the best regu-lation and supervision (chapter 4 will delve into possibilities of import-ing financial services as a way to lessen this vulnerability) As suggestedbelow, regulation and supervision, with some notable exceptions, arenot the strongest there.

Structural issues canmake emerging markets

more vulnerable tofinancial crises—

Figure 2.3 Volatility by region, 1970–99

Note: The median of the historical standard deviations of GDP growth and inflation for eachgroup of countries is expressed as a multiple of that for industrialized countries.

Source: Caprio and Honohan (1999); International Financial Statistics.

Standard deviations as multiple of industrial countries

Africa

Latin America and the Caribbean

Middle East and North Africa

Other East Asia

"Asian Miracle"

Growth

Inflation

Growth

Inflation

4321

1 10 100 1,000

Developing countries displaygreater nominal and realvolatility, compared with

industrial economies.

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0486

87


Not surprisingly, greater economic volatility translates into financialmarkets. Although based on a few cases with a long availability of data,figure 2.4 shows not only that equities enjoy a far higher return than eitherbills or even bonds in emerging markets relative to that in high-incomecountries, but the differences in volatility are even more dramatic. Giventheir greater volatility, then, even if local banks diversified in emergingmarkets, or were equally well regulated, they would enjoy much less stabil-ity than banks in the safer haven of most high-income countries. Exchangerate volatility also has had marked consequences in developing economies

Figure 2.4 Volatility in asset markets

Source: International Financial Statistics; IFC Emerging Markets database.

Percent per annum

Mean real return by asset

India Republicof

KoreaThailand United

States

BillsBonds

Equities

20

10

0

Percent per annum

India Republicof

KoreaThailand United

States

BillsBonds

Equities

Standard deviation of real return by asset

50

25

0

Stocks can pay better inemerging markets...

...but are riskier still.

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0487


88

because they largely have had to borrow in foreign exchange. Thus, in-creases in dollar interest rates often induce a larger increase in domesticlending rates, to the extent that the currency risk premium rises. Thisadditional volatility affects firms and their financiers. Chapter 4 returns tothis theme of volatility and small financial systems.

Third, emerging financial markets are dominated by banks (figure 3in the overview), meaning more demandable debt, higher debt-to-equity ratios, possibly inducing greater fragility. If a firm is 100 percentfinanced by debt, then even a small shock that reduces its projectedrevenues or raises its interest cost can result in the firm’s becoming insol-vent. Equity acts like a buffer, providing the firm with greater flexibilityin comparison with the need to service fixed debt repayments. Highdebt-to-equity ratios were found to be a factor in the East Asian crises;although these ratios did not in general increase in the immediate run-up to the crisis, their high level meant that the firms and the economywere highly fragile (Claessens, Djankov, and Xu 2000).

Similarly, if firms can only obtain financing that has to be renewedfrequently—every 90 days or more often—they are in a less flexible po-sition to deal with unanticipated shocks, compared to those with a highermix of long-term debt. Thus, the relative underdevelopment of non-bank finance and capital markets means that when developing countrybanks get into difficulty, the impact on the entire financial sector andthe economy is greater than in industrial countries, where nonbank in-termediaries and markets are generally better developed. More financ-ing through equity-type instruments transfers the risk to those morewilling and able to accept it. Availability of equity finance thus repre-sents an important potential buffer for the finance of firms, and indi-rectly for their bankers. The equity market can be seen as a spare tire forfinance (Greenspan 1999). Collapses in equity prices are not innocuous,but are clearly less disruptive than bank failures—which is why this chap-ter focuses on the latter.

Unbalanced financial systems with bank dominance are in part a re-sponse to the greater information problems in developing markets—hence the importance of improving this part of the sector’s infrastruc-ture (chapter 1)—but also likely reflect excessive “subsidization” ofbanking through the safety net (described below) or state ownership,which provides an implicit safety net for all bank creditors. State owner-ship itself appears linked to fragility (chapter 3).

—including the dominationof banks in the financial

systems—

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0488

89


Fourth, in addition to short-term volatility, there have been a succes-sion of regime shifts altering the risk profile of the operating environ-ment in hard-to-evaluate ways, including most prominently financialderegulation. In line with prevailing intellectual trends and followingthe example of industrial countries, emerging market authorities removedor eased administrative controls on interest rates, bank-by-bank creditceilings, rules for the allocation of credit to preferred sectors or borrow-ers, limits on new entry, and even opening the capital account. Disman-tling many old controls would ultimately have become inevitable, butacademics, advisers, and policy officials alike failed to realize the com-plexity of the task they had undertaken.

The enthusiasm with which liberalization was adopted in some coun-tries in the absence of necessary institutional underpinnings left finan-cial systems facing largely uncharted territory. New owners and inexpe-rienced bank supervisors tried to feel their way to an assessment of whatsafe-and-sound banking would mean in practice. At a minimum, thissituation suggests a fifth factor behind emerging market crises, namely aregulatory and incentive environment ill prepared for a market-basedfinancial system, and in particular one that encouraged or condonedexcessive risk-taking.

Poor sequencing of financial liberalization in a poorly prepared envi-ronment has undoubtedly contributed to bank insolvency. Countriesabandoned controls on bank liabilities—notably interest rates—but thetime to create and implement oversight of assets was greatly underesti-mated. Only if institutional underpinnings are strong is financial liber-alization unlikely to add to the risk of systemic bank failures (Demirgüç-Kunt and Detragiache 1999). It would be misleading, however, toconclude that greater reliance on market forces was always the underly-ing source of bank failure. In many cases, financial liberalization hasrevealed a long-standing underlying insolvency of the banking system,which became unavoidably clear as the banks emerged from the shel-tered environment that allowed or required them to cross-subsidize loss-making lines of business.

Authorities did not liberalize finance in a vacuum, but rather as part ofa general move away from heavier government intervention. The struc-tural economic transformation in many transitional and developing coun-tries created a new economic and political landscape and placed bankersin a brave new world with a shortage of skills and experience for judging

—and the poor sequencingof financial liberalization

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0489


90

the level of risk. With all these changes, in addition to those entailed bythe revolution in technology, communications, and financial engineering,plus the seemingly fickle behavior of international investors, it is hard forbankers, governments, and regulators to judge what sources of volatilityare likely to be important, and thus what constitutes sound banking.

These factors behind emerging market crises suggest first that, whilemoving in the direction of the market-based regulatory framework mayhelp, the special factors that characterize these economies necessitate evenmore robust measures.

Regulating Banks: Harnessing the Market

FOR AS LONG AS THERE HAVE BEEN BANKS, THERE ALSO HAVE

been governments setting a number of rules for them, such asmaintaining the purity of coinage and regulating exchange at

medieval fairs, holding high, even 100 percent reserves (in 16th centuryEurope and later in U.S. banks), maintaining interest rates below usuriouslevels, and providing credit to the ruler, especially in times of war. Modernfinancial regulation includes an array of instruments designed to improvethe informational efficiency of financial markets, protect consumersagainst fraud and malfeasance, and preserve systemic stability.7 Prudentialregulation promotes systemic stability. Whether or not there is a depositinsurance scheme, the official prudential supervisors in effect act asdelegated monitors for depositors, exploiting economies of scale toovercome information problems that would be beyond the resources ofsmall depositors.

Many proposed rules for reducing banking risk look promising atfirst sight, but prove to have serious drawbacks and can only be recom-mended, if at all, where all else has failed. One recurring example isthe idea of narrow banking, a proposal with a lengthy history (box2.2). It amounts to saying that, given the particular fragility of theliability structure of banking, why not make banks safe by forcing themto hold safe assets? As with many recommendations for finance, so-called narrow banking plans may fit some countries, such as those thatfollowing a crisis have banks with balance sheets dominated by gov-ernment paper. Although these plans in general have merit, they donot address the need for intermediaries to intermediate risk, the act ofwhich can create a problem when it goes wrong, but which can be anenormous benefit to growth when done well. If narrow banks hold

Prudential regulationpromotes systemic stability

Narrow banking couldthrow the baby

(intermediation) out withthe bathwater (crisis)

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0490

91


safe assets, but other intermediaries finance risky investment, the lat-ter will pay higher interest rates, and if the history of finance is anyguide, almost certainly attract many depositors, eventually make losses,and eventually mount lobbies for government protection.

BANKS THAT TAKE DEPOSITS AND DO NOT MAKE

loans are not new, and with the originalgoldsmiths—those who guarded depositors’ gold—being the earliest example of “100 percent reserve”banks. As bankers learned that not all depositorswanted their funds returned simultaneously, theybegan to lend out part, embarking on fractionalreserve banking, but a number of countries had orstill have banks that mainly hold safe instruments.In 1864 the U.S. National Bank Act required note-issuing banks to hold $111.11 in government bondsfor every $100 of notes issued, and this systemremained in force until the 1930s. Similarly, postalsavings banks in many now industrial economiesand some ordinary saving institutions (such as inFrance and the United Kingdom) required thatdeposits be invested in government paper. In thesecases, however, 100 percent reserve banks were onlypart of the banking sector, and other banks wouldtake deposits and make loans.

The Depression in the United States and in par-ticular the extreme panic in early 1933, culminatingin the banking holiday of March 1933, led to pro-posals by Henry Simons and a number of other pres-tigious economists for a 100 percent reserve bank-ing plan as the model for the country. Bankingproblems regularly unearth new interest in this pro-posal, as seen during the U.S. Savings and Loan cri-sis and in Argentina in the 1995 crisis.

The basic plan is simple: if all banks hold only de-posits backed by high-grade instruments, such as short-term treasury bills, perhaps even quite high-grade com-mercial paper, the payments mechanism will beprotected (except from a run on the currency, which

can be averted only if sufficient reserves are denomi-nated in foreign currency). As is the case with U.S.money market mutual funds, failure can only occurbecause of fraud, which has not occurred on any butan isolated basis. Other financial intermediaries, orthe nonbank subsidiary of a financial conglomerate,according to these plans, would be allowed to lend,but they could not call themselves banks, and theywould not be eligible for any deposit insurance. Thus,the goal is to attempt to convince depositors that ifthey want a guaranteed return, it will be a low one,and that funds placed in risky investments can be lost.

The history of finance suggests that plans wouldbe evaded. Thus, the U.S. National Bank Act wasmade less effective as banks began to issue liabilitiesthat were not reservable, and therefore yielded pro-ceeds that could be lent out profitably. Also, plans toencourage excessively easy financing of governmentdeficits could encourage excessive borrowing, in par-ticular in countries with inadequate fiscal controlsand established checks and balances in government.Transition to narrow banking could be tricky and, asnoted in the text, the fundamental problem of inter-mediation would remain.

Still, narrow banking might be suitable for somecountries as part of crisis response. For example, incountries where all or most of the banks have hadlarge parts of their assets replaced with governmentfunds, these banks already are virtually narrow banks,and a separate institution could be licensed to makeloans. Some regulations would be needed to encour-age transparency of the nonbanks, and an educationcampaign would be required to ensure that deposi-tors were aware of their exposures.

Box 2.2 Narrow banking

Source: Phillips (1995).

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0491


92

Thus, for most countries, it seems safe to assume that narrow bank-ing will not solve the fragility problem. Moreover, there may well be atradeoff between stability and efficiency. If the formation of narrow banksdid not lead to a large migration of assets to nonbank intermediaries, theallocation of resources to efficient investments might be seriously im-peded. Although banking has declined some in relative importance inadvanced countries, it remains significant and in developing countries isthe dominant portion of the financial sector.

Although small investors can suffer losses in nonbank finance, too,(and official safety nets are sometimes provided to consumers in seg-ments of the insurance and pensions fund industries), failures and lossesin financial markets that do not extend to the banking system are muchless likely to have catastrophic systemic effects on the payments and creditsystem.8 For this reason, nonbank financial intermediaries and marketsare also objects of generally lighter government regulation—from thegreater oversight in pensions and insurance to less oversight in stocks,futures, and derivatives markets.

Financial sector regulation and supervision—the rules of the game inthe financial sector and the way they are enforced—are essential to limitmoral hazard, as well as to ensure that intermediaries have the incentive toallocate resources and perform their other functions prudently. In the 1980sand 1990s, many developing countries began making the transition awayfrom supervisory systems aimed at ensuring compliance with governmentdirectives, such as directed credit guidelines and other portfolio require-ments and toward what might be called the Basel standard, which is oneof supervised capital adequacy. As noted earlier, this transition has notgone smoothly, and evidence suggests that liberalization, at least as con-ducted, even contributed to the recent spate of banking crises.9

In response to these crises, there has been a boom in the creation ofdetailed standards that are being promulgated in banking (and other ar-eas of the financial sector). These standards may ultimately induce im-provements in the regulatory environment, but the absence of a clearsense of their relative importance or how they function in the disparateinstitutional contexts found in emerging markets reduces their impact.The outcome of research on financial systems, on the other hand, sug-gests that rather than a large number of standards, authorities in emerg-ing markets should focus on using incentives to harness market forcesthat favor effective and efficient financial markets, and employ individualstandards to the extent that they contribute to this purpose. To someextent, this means imposing tough rules—not only requiring minimum

The transition to modernprudential regulation is

difficult

Authorities should useincentives to harness

market forces

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0492

93


capital ratios, but perhaps more robust restraints, such as minimum di-versification guidelines (or tailoring capital requirements to the concen-tration in banks’ portfolios) or requiring a certain proportion of the bank’sliabilities to be in the form of uninsured subordinated bonds. The degreeto which the authorities can use such rules to exploit market informationand market discipline depends to some extent on the level of overall fi-nancial market development.10 This section examines the extent of regu-latory convergence between developed and emerging markets, includingthe problems of applying regulatory choices in the former to the latter,and then focuses on how the market can best be harnessed to help pro-duce safe and sound finance.

Although there has been a remarkable convergence on paper in re-cent years, stark differences remain in the regulatory environments aroundthe world. Thus, at the time of the 1988 Basel Accord, which recom-mended a minimum risk weighted capital adequacy ratio of 8 percent,some developing countries did not even have capital requirements, andmany that did had low ratios (2–5 percent not being uncommon) anddid not engage in prudential supervision to verify them. By 1998–99, of103 countries reporting, only 7 had minimum capital ratios under 8percent, and 29 had minimum capital ratios of 10 percent or more, onlyone of which was from the OECD region. And more than 93 percent ofall countries (88 percent in emerging markets) claim to adjust capitalratios for risk in line with Basel guidelines.

It is easier, however, to adopt “headline” regulations, such as capitaladequacy ratios, but more difficult to implement the underlying proce-dures and to acquire the necessary supervisory skills to give teeth to theserules. Unfortunately, capital by itself is an inadequate indicator of thehealth of a bank. The true net worth of a bank depends on the quality ofits portfolio which, for many banks, is dominated by illiquid loans thatcannot easily be valued or “marked to market.” This problem is all themore real in developing countries, where volatile prices and thin or non-existent markets render such estimates hazardous. All too often a bank istruly insolvent long before its accounts tell us so. If capital is actuallynegative, risk adjustment is irrelevant.

What matters for true net worth is capital net of provisions for loanlosses, but accounting rules in many countries permit bankers to beoptimistic and underprovision. If the bank has reached a reasonablemeasured capital adequacy ratio only because it made no provisionsagainst loan loss (P = 0 in Table 2.2), we can safely say that its truecapital is below standard. Even an insolvent bank (with a true P of 10

A convergence of headlineregulations—

—notably the accountingfor loan loss provisions

—but wide disparities intheir effectiveenforcement—

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0493


94

or more) can remain in business for months or even years, provided itdoes not run out of cash. As long as the net inflow of deposits and theinterest received on performing loans are sufficient to pay operatingexpenses and interest on deposits, closure can be deferred. Depositorsand supervisors may be lulled into a false sense of security if account-ing rules are flouted. Accounting rules in some countries still havesome way to catch up here.

Rather than rely on historic values, bank supervisors classify loansinto forward-looking categories, such as “normal,” “specially mentioned,”“substandard,” “doubtful,” and “loss,” and regulations implicitly attachloss probabilities to each of the last three categories by requiring a cer-tain percentage (typically 20, 50, and 100, respectively, in the latter 3categories) of the value of loans to be provisioned in the bank’s accounts(usually in addition to some general loan-loss provision of 1 or 2 percentof the entire portfolio). Indeed, here too our survey shows that require-ments are on average slightly tougher on paper in low-income countries.What is important here, though, is that the provisioning requirementsshould actually correspond to subsequent loan-loss experience.11

Unfortunately, ensuring adequate, forward-looking classification of loansis not straightforward. Especially when economic conditions move out ofthe normal, or for the large or unusual loans that are often the weak pointof a reckless bank, experience may be a poor guide, even to the banker.The high-risk environment and rapidly evolving economic structure ofmost developing countries obviously exacerbate the severity of this prob-lem. Realistically, in the face of a resistant bank management, given theinherent difficulty in understanding the true risks, supervisors often can

Table 2.2 Typical balance sheet

Assets Liabilities

Cash 10 Demand deposits 100Liquid investments 20 Other debt 30Loans at historical value 100Less provision for loan losses –PProperty 10 Capital 10–P

Source: ???

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0494

95


do little more than rely on a backward-looking measure: insisting on pro-visions being made when the loan goes into arrears. In this respect, theaccounting rules or standards vary widely. In particular, low-income coun-tries typically are more lenient than the upper-middle-income group (fig-ure 2.5). Also telling is that one in three low-income countries allow banksto treat interest that is in arrears as earned income, at least for a time. InThailand interest accrual on nonperforming loans was allowed for up to360 days in 1997 and for 180 days in many African countries. In mostcountries it is still more difficult to prevent a bank from concealing anonperforming loan simply by “evergreening,” that is, by making a newloan to cover the repayment. Most tellingly, Cavallo and Majnoni (2000)show that whereas industrial countries build up provisions in good timesand draw them down as the business cycle weakens, there was no suchvariation in the developing countries in their sample, again suggestive thatconvergence to industrial country norms is more superficial than real.

In sum, measuring the size of the buffer is a challenge that is far frombeing under control. Although not published, the Basel Core Principleassessments are understood to be revealing that developing countries areconsiderably further from full compliance than their industrial countrycounterparts. Headline regulations are promulgated without having the

Figure 2.5 Classification of substandard loans, 1997

Source: Barth, Caprio, and Levine database.

Number of days till a loan in arrears is substandard Number of days till a loan in arrears is doubtful

120

80

40

0

250

200

150

100

50

0Low

incomecountries

Lowmiddleincome

countries

Uppermiddleincome

countries

Lowincome

countries

Lowmiddleincome

countries

Uppermiddleincome

countries

Less rigorous loan classificationstandards apply in lower incomecountries.

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0495


96

information needed for verification or without putting in place the in-centives that might help reveal it.

One should not be dismissive of the ability of official supervisors touncover problems. Empirical evidence exists that they can and do pro-vide independent information. For example, Jordan, Peek, and Rosengren(1999) found for the United States that the release of adverse supervi-sory information resulted on average in a 5 percent decrease in the bank’sstock price, suggesting that the release did contain news. Not surpris-ingly, there was some variation. Banks that had already disclosed badnews saw little effect, and there was little evidence of contagion, in thesense of other banks’ stock prices reacting when another bank disclosedinformation, except in the case of a common, regional shock for banksin the same region.

This evidence shows not only that good supervision can have an ef-fect in that it does reveal additional information and can lead to theissuance of supervisory actions designed to stop imprudent behavior. Italso points to the advantages of greater disclosure in that markets canpressure banks to adjust as soon as possible and before a crisis results.

How does one get good supervision? The Basel Committee guide-lines provide supervisors’ views on this, and there is little doubt thatfactors, such as the independence of the supervisory agency, are key togood supervision.12 Here we note the issues related to the incentives thatsupervisors face.

It must be recognized that the environment in which prudential regu-lation and supervision is being conducted differs markedly between in-dustrial and developing countries. In addition to the greater volatility ofemerging markets, income and wealth tend to be much more highlyconcentrated than in industrial countries, and recent evidence showsthat this holds for the ownership of corporations as well (figure 1.11). Itis not hard to see that this adds to the challenges faced by supervisors, byincreasing the likelihood that the financial firms under their supervisionare controlled by extremely powerful individuals.

The result can be a skewing of the “balance of terror”—the risks andrewards faced by official supervisors in many countries. First, supervisorsgenerally are paid less well relative to salaries in private banks, and inmany developing countries turnover is becoming even more of a problemthan in industrial countries. Second, deferred income—a potential bo-nus, in effect—can result from lax supervision, since only a few coun-tries, regardless of income level, have prohibitions on supervisors movingto work for banks. Third, there is no deferred penalty—neither through

Good supervision canimprove the health of the

financial system—

—but incentive structuresoften make this difficult

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0496

97


a loss of bonuses, which generally are not offered—nor by a forfeiting ofpensions. And last, in several countries well into the middle-income range,such as Argentina and the Philippines, supervisors can be sued for theiractions and be held personally liable, so they face a very real penalty nowfor vigorous action.

This is precisely the opposite of the optimal compensation structurefor those charged with enforcing laws and regulations that has long beenrecommended for eliminating malfeasance even when it is difficult todetect bad conduct.13 So, a priority for securing better supervision is topay bank supervisors well, even by reference to other public servants: theprobability of detection of malfeasance is low and, as seen in figure 2.1above, the cost of laxity on their part is high. Given that it may takesome time for supervisory laxity to be evinced, deferred compensationwould be the best way to motivate supervisors. Thus, providing themwith a generous pension as a deferred bonus, and then removing or re-ducing that pension for violations of good supervisory practice will helpimprove incentives. In addition to the common view that supervisoryagencies require a high degree of independence to reduce political inter-ference, if supervisors were simultaneously protected against personalliability (as in many industrial countries), more countries would be ableto benefit from more vigorous enforcement.

Transparency and accountability alone are not sufficient for bettersupervision. This approach may be sufficient, for example, to ensurethat central bank governors behave responsibly in setting monetary policy,because exchange rate and/or bond markets provide a ready assessmentof their actions. Also, most central bank governors do not face lawsuitsfor tightening policy, nor are they rewarded in the future for lax policy.Although the reaction to the U.S. savings and loan problems was toreduce supervisory discretion—through mandatory, prompt, correctiveactions—the growing difficulty because of the plethora of financial in-struments in observing the risk position of banks is leading to morediscretion for supervisors, for example, by having them agree with bankson how they model risk and then penalizing them for violating the model.This is not an area easy to monitor. To the extent that developing coun-try supervisors move in the same direction, it will be particularly impor-tant that greater discretion is accompanied by greater oversight and acorrected balance of terror.

Although it is necessary in many countries to improve supervisorycompensation, it is both unlikely and costly to pay supervisors salariesthat are equivalent to senior bank officers. Forcing greater revelation of

Correct the “balance ofterror”—

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0497


98

information is the standard way to limit the required increase in theefficiency wage, so something like the above subordinated debt proposalis especially important to force greater disclosure of market informationand sentiment.

Although the financial conditions of banks are difficult to assess evenin industrial countries, the above suggests that it is especially risky inemerging markets to put excessive reliance on official supervision. Therecurrence of fraud, defalcations, and crises demonstrates that the infor-mation and incentive problems that dominate finance are not easily elimi-nated. Moreover, differences in institutional development and economicvolatility, combined with the ability of financial market participants toadjust to regulation, mean that rather than precise forms or rules, au-thorities need a strategy for approaching financial sector regulation, andthe strategy has to go considerably beyond convergence to industrialcountry norms.

With greater income and ownership concentration, it is more diffi-cult to maintain adequate independence of supervisory agencies. Also,the information environment, degree of public oversight of supervisors(not just disclosure, but the degree of sophistication of the press on fi-nancial matters), and the basic incentives that supervisors confront allwill operate to yield less effective supervision. Political interference inbank supervision has happened even with good checks and balances,such as in the United States as savings and loans had members of Con-gress lobby for lighter regulation and reduce regulatory capital require-ments. These potential problems are likely to be more pronounced whereownership concentration is greater (for example, Venezuela in the early1990s, in which a senior central bank official owned shares in a bank).

Besides, just as authorities in developing countries were making thetransition to supervised capital adequacy, the goal posts were moving.First, the complexity of modern finance has amplified the difficulty ofsupervising on transaction-by-transaction basis. In part, with the growthof derivative instruments, banks can now shift their exposure withinminutes, so that reviews of their current exposures convey less informa-tion as to their health than they would have previously. As already men-tioned, this has led middle- and upper-income countries, where suchinstruments are more prevalent, to shift the focus of supervision to thebank’s risk management systems, though experience with this approachis still limited.

—and supplement officialsupervision with market-

based monitors

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0498

99


Second, as noted above, banks are adept at adjusting to a set of rules.The arbitrary risk weights of the 1988 Basel Accord were easy to evade,and indeed sparked a decade of financial innovation at least in part withthis purpose in mind.

The answer from recent and historical research on financial systems isremarkably clear, though as just seen, not always as simple as it appears:use incentives and information to maximize the number of well-informed,well-motivated monitors of financial intermediaries.

Understandably, diversity in the set of monitors for banks is desirablenot only because of possible differences in the information they maypossess, but also reflecting the varying and possibly opaque incentivesthey face. Who else, though, apart from official supervisors, can moni-tor banks? Three classes of monitors should be considered:

• Insiders, including the owners, the board, and senior managementof a bank, whose net worth should, in an ideal world, depend onthe prudent performance of the institution.

• Rating agencies.• Markets, meaning all nonofficial outside creditors and

counterparties.

Owners earn returns on the capital they have invested. These rewardswill be based on current and expected future profits, or the so-calledfranchise value. Profits in turn will derive from the regulatory frame-work that constrains banks to various activities and ways of doing busi-ness. If the profits from prudent banking are high, and if the threat thatbanks could lose their bank license (and thus their equity and the relatedrewards) is real, owners will be motivated to preserve their franchise value.Majority owners and senior managers may be in the best position tosurmount information problems, but as numerous bank failures show,such as the famous 1995 Barings episode, owners of large, complicatedintermediaries still face these problems. Minority owners do not neces-sarily have any better information than the general public.

Bank directors have the responsibility of representing all owners, andof disclosing accurate and timely information on their institution. Bet-ter and more timely information will improve the ability of all outsidersto monitor them. Most countries in theory make bank directors respon-sible for accurate disclosure, but in only a third do they have enforcedpenalties, most of which are in high-income economies. Enforcement is

Use the private sector toextend the reach of theregulator

Banks often reward risk-taking

02 Chap02 Ed.1a (noRed) 3/19/01, 9:0499


100

critical. Stiff penalties for inadequate disclosure, and more generally forexcessive risk-taking, is a way of increasing the liability of owners be-yond just the capital they have invested for the performance of the bank.

If bank directors and majority owners were highly motivated to en-gage only in safe and sound banking, they would likely endeavor toeffect a compensation system for senior bank management that wouldreward prudence. However, the fallout from the Long Term CapitalManagement (LTCM) fiasco revealed that senior executives of a few largeinternational banks were forced out—the good news—for making simi-lar bets as LTCM, but—the bad news—they were able to takemultimillion-dollar bonuses with them. In all likelihood, this reflectsthe predominance of banks that are willing to gamble and hence offercompensation packages that attract risk takers.14 Authorities could try tocorrect for this market failure by making capital ratios or deposit insur-ance premia a function of the compensation structure for senior man-agement. Supervisors in many advanced economies do look at risk man-agement systems that banks have and grade them on this effort. Thesuggestion here is that the source of the risk management system, execu-tive compensation, rather than its advertised manifestation, be factoredin to regulation. The compensation structure also should be disclosed—not just the raw salary, but how bonuses and other forms of compensa-tion are determined (John, Saunders, and Senbet forthcoming).

One recent proposal for bringing the views of private market partici-pants on bank risk to bear was advanced in 1999 and 2001 Basel Com-mittee discussion papers seeking to reduce the arbitrariness of the riskweights attached to bank capital requirements by proposing that the weightsinstead be derived from ratings publicized by approved external credit as-sessors (e.g., rating agencies). Although this proposal would appear to bean attempt to “harness the market,” it is instructive to consider severalproblems facing implementation of this proposal, especially in developingcountries. Among the better known difficulties are the following:

• It is unclear how reliable rating agencies would be where informa-tion costs are high, the ratings industry is at best nascent, and wherebanks often pay for their own ratings.

• Ratings are based on expected default rates, but capital is intendedfor unexpected losses.

• Ratings for tradable bonds face issuers who want favorable ratingsand investors who want protection, whereas banks and their

The problems with ratingagencies

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04100

101


borrowers may both want to secure a favorable rating. Will ratingsreally be done at arm’s length, especially where ownership and in-come are concentrated?

In addition, however, are a number of less-recognized points that arehighlighted by a focus on incentives (Honohan 2001b). First, the usualmoral hazard problem will be exacerbated. If it is announced that bankswill have to hold capital in accordance with the riskiness of their portfo-lio, each borrower will have the incentive to secure a favorable rating,even though it continues to place the bank—and the deposit insurancefund, if one exists—at risk. Bankers, assuming that they have decided tomake the loan, will be motivated to collude or go along with a favorabledepiction of their borrowers, because it will give them greater freedomin making capital decisions.

Second, raters may release less information about borrowers so as notto lose business. And most serious of all, rating agencies are not paid toanticipate the risk of correlated, systemic shocks, so even if the averagerating of a borrower is accurate for normal times, it will not be for acrisis. This problem is especially serious as developing country authoritiesmay believe that by using (“market based”) ratings, they are protected againstcrises, when in fact they are not. Even though rating agencies in the UnitedStates do a fair job on individual firm ratings, their ratings perform lesswell on emerging market paper precisely because it is difficult to esti-mate systemic shocks in small, volatile economies.

Thus, it is important for authorities to use market forces, but this dis-cussion illustrates that it is equally important to understand what the in-centives are and how they operate. Also, rather than worry about how tomotivate rating agencies to take proper account of correlated factors, au-thorities should focus on banks, which can and should be looking at theirentire portfolio and how it varies or is exposed to different risks. Compel-ling banks to disclose certain information can be part of this process sothat agents external to the bank who have the right incentives will put thisinformation to good use. Relying on rating agencies puts excessive burdenon entities that may not have as much to lose as bank creditors do.

Given the incentive that equity holders and other insiders may haveto increase risk, and the uncertainties of relying on rating agencies, it isall the more important to consider how the incentives of other bankcreditors can be aligned with the social goal of limiting bank risk. Al-though small depositors may choose to “free ride” on other claimants,

Outside creditors can actas monitors—

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04101


102

large creditors, if they have no expectation that they will be compensated fortheir losses, have clear incentives to monitor banks. Recent proposals at-tempt to capitalize on this incentive by forcing banks to issue subordi-nated debt, that is, a fixed claim that is only senior to equity. Not enjoy-ing the upside gains of equity holders, but holding almost as much ofthe downside risk, subordinated debt holders would be highly moti-vated to police banks for excessive risk-taking. Also, they would not botherwith a “loan-by-loan” analysis that is part of the current Basel Commit-tee process, but rather be concerned with the overall risk that banks face.Other large creditors—such as other banks in interbank markets—wouldalso be motivated to monitor banks as well, as long as they were notunder the presumption that they might be “bailed out” if the bank gotinto difficulties.

Subordinated debt is not new—as of 1999, 92 of 106 countries re-sponding note that they allow subordinated debt to fulfill some part oftheir capital requirement. However, those countries in compliance withthe Basel Committee guidelines in effect regard it as cheap equity, andto that extent only make eligible long-term debt, and then limit the useof such debt. This, however, ensures that rollovers of the debt will berelatively rare. Also, the fact that it is not required to be issued and is notpoliced then leads to its issuance to firms that are not at arm’s length. Yetregular issuance, tradability, and arm’s-length issuance all are needed toensure better monitoring. To prevent this debt from becoming a kind of“junk bond,” it will be necessary to put some cap on the interest ratethat can be paid. If these features are present, subordinated debt holderswill be even more concerned to avoid a bank that is taking imprudentrisks than are the supervisors (and taxpayers). Far from being cheap eq-uity, this kind of subordinated debt can be a valuable discipline. There ismuch to be said for requiring its issuance, especially for larger banks ineach country. To provide reliable monitoring, subordinated debt holderswould become an important lobby group to press for a number of theimprovements to infrastructure and information noted earlier, particu-larly related to the disclosure of information.

To be sure, subordinated debt proposals (box 2.3) can be quite diffi-cult to implement. Capital markets in developing countries are thin,though a requirement that banks issue this debt would deepen them some-what. Most importantly, a key to its success is to ensure that the issuersare truly at arm’s length from the holders of the debt, meaning that they

—and a subordinated debtrequirement is a

promising, but not fool-proof, way to improve

market monitoring

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04102

103


SUBORDINATED DEBT CAN SERVE AS A BUFFER TO

absorb losses, but probably its most valuablecontribution is by the signals it can provide as to bankriskiness. This signal both will serve as a discipline inthe market, as banks find it harder to renew theirsubordinated debt or find the interest rate thereon risingas risk increases, but also by the indirect signal it providesto others, including bank supervisors. The latter benefitcould be great. One problem with so-called prompt,corrective action proposals is that the criteria forintervention still leave significant responsibility tosupervisors, which may be particularly difficult incountries in which the institutional independence ofthe supervisory agency is in doubt. A recent study ofthe Board of Governors of the Federal Reserve System(1999) noted that one difficulty for officialsupervisors—the burden to prove that banks may betaking excessive risks—does not hold for subordinateddebt holders, who instead get to place the burden ofproof on bank managers who need funding. Supervisorscould use either the interest rates or ability to issuesubordinated debt as a signal to increase monitoring ofrisky banks or to take mandated actions, or both.

How should it be issued? A requirement thatbanks issue this debt regularly in ‘lumpy’ and rela-tively homogeneous forms would produce a well-informed monitoring system for banks; the regularissuance would continually “refresh” market infor-mation, in that banks would presumably find it ad-vantageous, and markets likely require, current in-formation at the time of issuance. If the subordinateddebt instrument is relatively homogeneous, then therate at which it trades could be more easily com-pared across banks, thereby facilitating monitoring.

In addition to tradability, maturity matters, andthe balance of opinion appears to be weighted to themedium term of 2–5 years. While Federal ReserveSystem interviews with U.S. market participants sug-gested that market depth would be greater with 3–5year maturity, Calomiris has proposed for emergingmarkets as well that banks be required to issue 2 per-cent of their nonreserve assets (or 2 percent of risk-weighted assets) on a monthly basis with 2-year ma-turity, so that every month they would have torefinance 1/24th of this debt. Calomiris (1999) alsonotes that banks in trouble could pay higher interestrates, but he would limit this by imposing an inter-est rate cap. That would mean that highly risky bankswould be forced to shrink the asset side of their bal-ance sheet and eventually close or otherwise restruc-ture their operations when they could not complywith the subordinated debt requirement

Whereas regular issuance would impose disciplineon issuers, there is a tradeoff between this gain andthe cost to banks—and their customers—from morefrequent and smaller issues, because of transactioncosts. Indeed, very small banks in emerging marketslikely could not pay these costs, so Calomiris has rec-ommended that small banks be allowed to satisfy asubordinated debt requirement by “issuing” large de-posits to a qualified institution. Because it is the largerbanks whose stability is essential for the health of theoverall system, and for which early intervention isimportant, this limitation is not likely to be severe.Last, to increase the likelihood that subordinate debtholders will be at arm’s length from the issuing banks,it may be necessary to put restrictions in place thatcould limit the attractiveness of this paper.

Box 2.3 Subordinated debt proposals

Source: Board of Governors of the Federal Reserve System (1999); Calomiris (1999); Evanoff and Wall (2000).

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04103


104

neither should be related parties, nor should the issuer be allowed toprovide comfort or guarantees to the holders. Ensuring this is not a trivialconcern, and is an excellent reason for not relying exclusively on subordi-nated debt holders to ensure safety and soundness. Greater reliance onsubordinated debt and on other uninsured creditors’ monitoring, how-ever, seems to be a worthwhile initiative in middle-income countries.

Notwithstanding the difficulty of ensuring arm’s length between banksand the holders of subordinated debt, early results from Argentina arepromising. Even though subordinated debt only began to be requiredthere in 1998, and though its implementation was delayed by the EastAsian crisis, banks that were largely compliant saw lower deposit rates,faster growth in deposits, a lower capital ratio, and a substantially lowerratio of nonperforming loans compared with noncompliant banks (fig-ure 2.6). More formal econometric analysis confirms that the subordi-nated debt requirement there has encouraged better monitoring andgreater prudence in risk management (Calomiris and Powell 2000). Evenif only good banks were able to issue subordinated debt there, this factof itself conveys important information to supervisors. The above evi-dence that credibly uninsured creditors are more likely to provide moni-toring of banks strengthens the promise of subordinated debt in im-proving the market monitoring of banks (Evanoff and Wall 2000). Again,however, it is important to stress that subordinated debt should not bethought of as a single cure for unsafe banking, but rather as a potentialtool in the regulatory arsenal.

Banks that complied withsubordinated debt requirements

paid lower deposit rates butenjoyed faster deposit growth, alower capital ratio, and a lower

rate of nonperforming loans.

Figure 2.6 Subordinated debt in Argentina, 1996–99

Source: Calomiris and Powell (2000).

Nonperformingloans

Capitalratio

Depositgrowth

Depositrate

Percent

30

20

10

0

High compliance

Low compliance

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04104

105


Financial Sector Safety Nets

I N THE FACE OF BANKING FRAGILITY, IT IS NATURAL FOR

depositors to hope for redress from government when things gowrong, but this expectation in itself can contribute to the fragility.

Although governments have a variety of mechanisms, such as thecentral bank discount window and other lender-of-last-resort (LOLR)facilities, which can be employed as part of a safety net for banks,explicit deposit insurance schemes are increasingly becoming a keycomponent, have an important impact on overall incentives, andtherefore are the focus of this section. Governments typically remainmore ambiguous about their LOLR function, which has been thesubject of an enormous literature.

Not surprisingly, deposit insurance arose where banking was mostfragile—U.S. states in which banking was conducted in unit banks(banks that were not permitted to branch) beginning with the N.Y.Safety Fund in 1829. Some 14 states (all with unit banks) adopteddeposit insurance; some failed shortly after their establishment, whileothers lasted until being done in during the agricultural collapse of the1920s. Only three systems—those that harnessed market forces—werejudged successful.

Still, by the late 1920s, the much better survival rate of branchingbanks appeared to have “won the day” for branching vs. unit banks (withor without deposit insurance) until the political realignment on this is-sue during the Depression. After the adoption of a national deposit in-surance system in the United States in 1934, the number of explicitsystems in other countries grew slowly for the first 30 years, with only 6being established, and then took off (figure 2.7).

Most deposit insurance systems are set up with either or both of thestated objectives of protecting the overall stability of the banking system,and protecting individual, especially small, depositors. In the pioneeringU.S. case, although political debate may cloud the true underlying pur-pose, scholars accept that it was systemic stability rather than small de-positor protection that was the key factor (Golembe 1960; see box 2.4).Other means of protecting small depositors were recognized, such as thesavings banks in Europe, which largely invested in safe instruments. TheU.S. deposit insurance legislation was passed by Congress in the midst ofthe banking crisis, though the run on banks—which was linked to fears ofdevaluation and other measures that might be adopted by the new admin-istration—had stopped before it went into effect.

Deposit insurance schemesare increasingly spreadingto emerging markets—

—with the goals ofprotecting the stability ofthe banking system, andthe savings of smalldepositors

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04105


106

More recently, some countries have adopted or expanded deposit in-surance during crises. For example, after two crises in the 1980s, Argen-tina abandoned deposit insurance in 1992, only to adopt a system oflimited coverage in 1995 in response to the Tequila crisis. Thailand movedto blanket insurance in 1997, including coverage of deposits at financecompanies. Mexico is the first developing country recently to have putin place plans to reduce blanket coverage, following its experience withthe 1994 crisis, so experience with this transition is necessarily limitedamong emerging markets. The sharp increase in the 1990s resulted inpart from the spread of deposit insurance to transitional countries, and

Figure 2.7 Explicit deposit insurance systems: the rise of depositinsurance around the world, 1934–99

Source: Kane (2000).

1996

1994

1992

1988

1986

1984

1982

1979

1975

1971

1967

1963

19611934

1962

1966

1969

1974

1977

1980

1983

1985

1987

1989

1993

1995

0 20 40 8060

Cumulative frequency of explicitdeposit insurance systems

established

1999

19971998

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04106

107


to some African states, perhaps reflecting the prevailing wisdom thatdeposit insurance would lead to a safer financial system.15

The systems that countries adopted differed dramatically. As men-tioned, some countries cover all deposits—including interbank andforeign currency deposits—and are even generous in extending the

Deposit insurance was not a novel idea; it wasnot untried; protection of the small depositor,while important, was not its primary purpose;and, finally, it was the only important piece oflegislation during the New Deal’s famous “onehundred days” which was neither requested norsupported by the new administration. (Golembe1960, pp. 181–82)

FOLLOWING THE ESTABLISHMENT OF THE N.Y.

Safety Fund (1829–66), Vermont (1831–58) andMichigan (1836–42) established similar schemes. Allexperienced severe losses in the panic of 1837, NewYork then allowed free banking, and its safety fundwas done in as better banks switched to become freebanks and thereby avoid the losses associated withpoor (public sector) supervision and limited premia.Vermont and Michigan also saw failures in the panicsof 1857 and 1837, respectively, also because ofadverse selection and poor supervision. Indiana(1834–65), Ohio (1845–55) and Iowa (1858–66)established more incentive-compatible systems:restricted membership, unlimited mutual liability, allprivately administered, and with powers to restrictdividends and impose other restrictions and penaltieson member banks.

These cases could not have differed more fromthe first three systems, in that there were few failuresand, like the states with branching (but without de-posit insurance), weathered common shocks quitewell. Interestingly, coverage was broad, but with un-limited mutual liability, the greater the coverage, the

greater the liability and thus the stronger the incen-tives to police one another. The systems ended withthe taxation imposed by the National Banking Sys-tem, and not because of crisis.

The post-Civil War period saw eight other statesadopt deposit insurance, and all perished with theagricultural crisis the 1920s, with the exception ofMississippi and South Dakota, whose schemes madeit until 1930 and 1931, respectively. So thoseschemes with mutual liability and private adminis-tration saw few failures, little or no evidence offraud, did not perish in crisis, and avoided suspen-sion during panics.

Rather than continuing to pay for the failures bythemselves, unit banking states regularly sought theprotection of the federal government, as 150 billsfor a federal deposit insurance system were introducedunsuccessfully between 1886 and 1933. Representa-tives from branching states continually opposed theattempt to make their voters pay for the fragility ofunit banking. The successful legislation was passedin 1933 after the bank run was ended by a bank holi-day and reopening of far fewer banks, but withoutincluding any ex post compensation for depositorsand with a low initial ceiling. Political compromiseappears to have been key: Carter Glass, chairman ofthe Senate Banking Committee and a long-time foeof deposit insurance, acceded to it as part of a dealwith Representative Henry Steagall to win passageof Glass’ plan for the eponymous banking act thatseparated commercial and investment banks. Glasslater said that the compromise was a mistake.

Box 2.4 The rise of deposit insurance?

Sources: Calomiris (1992), White (1997), and Golembe (1960).

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04107


108

coverage to a broad array of institutions. However, most deny—at leastin principle—coverage for interbank funds, so as to induce banks—who are large and supposed to be sophisticated, relative to many oth-ers—to monitor one another.16

Figure 2.8 shows the dramatic dispersion in the stated coverage of de-posit insurance relative to per capita GDP, for those countries with limitson coverage.17 Compared to the relatively modest protection in high-income countries, some of the poorest countries offer the most generousprotection, going well beyond the scale of the deposits of the poor—thoughthe extremely low level of average income in countries like Chad needs tobe kept in mind to put their coverage in perspective.18

Some deposit insurance schemes are funded or administered by theprivate sector, or both. And whereas many deposit insurance systems areprefunded, some 10 systems—mostly in Europe—as of 1999 were un-funded, with the power to make assessments on individual banks whenneeded. Most deposit insurance systems feature a flat premium, but abouta quarter feature some differential pricing, in effect an attempt to varythe premium with the riskiness of the individual bank, though the dif-ferential itself is small and not always collected.19

It is not hard to see why explicit deposit insurance systems have be-come increasingly popular. The political calculus is in their favor. Forone thing, they can appear to be a direct and seemingly costless solutionto the problem of bank panics and runs. Protection of small depositorsis also politically attractive. There are other political forces favoring theintroduction of deposit insurance, too. For example, a deposit insurancescheme can help small local banks in emerging markets acquire or retaintheir market share of deposits that might, in the absence of insurance,migrate to large and especially to foreign-owned banks.

Last, by providing a deposit insurance scheme, the government mayfeel that, in political terms, it is also buying the right to step in withregulatory intervention, as necessary, including the right to close un-sound or insolvent banks. This argument, however—that deposit in-surance is a necessary quid pro quo for the authority to close banks—goes too far. Almost everywhere in the past century, banking has notbeen a right, but a privilege, regulated by the state—and for good rea-son. Banking law properly requires licenses to be granted only to “fitand proper” individuals, and with the possibility that the license canbe revoked for improper actions, which should be defined as any thatviolate banking regulations.

Deposit insurance schemesare politically popular—

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04108

109


Deposit insurance coverage isrelatively generous in low-income countries.

Figure 2.8 Deposit insurance coverage

Note: For Germany only compulsory coverage is shown; the private voluntary systems have higherlimits, with each depositor protected up to about 30 percent of bank capital.

Source: Demirgüç-Kunt and Detragiache (2000).

Deposit insurance coverage/per capita GDP

Ratio for average OECD country

0 4 8 1612

Chad

FYR MacedoniaPoland

UkraineLatvia

EstoniaLuxembourgSwitzerland

LebanonDenmarkBelgiumBahrain

ChileIcelandIreland

GermanyNetherlands

AustriaHungary

SpainFinlandSweden

BulgariaGabon

United KingdomPortugalTanzania

Trinidad & TobagoSri Lanka

Republica Bolivariana de VenezuelaGreece

JamaicaRomania

NigeriaSlovak Republic

CanadaLithuaniaColombia

Czech RepublicEl Salvador

FrancePhilippines

Equatorial GuineaArgentina

United StatesCroatiaTaiwanBrazil

Republic of CongoKenya

ItalyIndia

BangladeshDominican Republic

NorwayUganda

PeruOman

CameroonCentral African Republic

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04109


110

The logic underlying the more persuasive political considerations isnot without merit. Credible deposit guarantees undoubtedly do forestallruns. Prompt repayment of their deposits is clearly a valuable protectionfor small depositors at failed banks, especially protecting them from infla-tionary erosion (though, as noted, there are other ways of offering safesavings media to low-income households, including postal savings banks—or even mutual funds restricted to secure money market assets). And ex-plicit deposit insurance does favor small banks, although if it comes at ahigh cost, governments would need to consider the tradeoffs carefully.

Less evident in the political arena, but long recognized by specialists, isthe fact that deposit insurance has the potential to induce greater risk-taking, or so-called moral hazard behavior. Limited liability allows bankowners to walk away from their losses—giving them the option to put thelosses to depositors or other parties. However, by reducing the incentive ofinsured depositors to monitor banks, deposit insurance can greatly ac-commodate risk-taking if accompanied by lax regulation and supervision.

Perhaps the most persuasive argument in favor of an explicit depositinsurance scheme is the thought that it can represent a limit to thegovernment’s commitment to depositors. Absence of an explicit systemmay really represent unlimited implicit coverage. By placing a modestlimit on the amount of deposit coverage, can the government effectivelysignal that it is not likely to indemnify depositors beyond this limit?

Clearly, the net impact of adopting an explicit system and, if so, ofimplementing various design features are empirical issues, and turn chieflyon the tradeoff between the gains from protecting depositors and thelosses from reduced market monitoring. Until recently, virtually no sys-tematic empirical research used data on emerging markets to addressthese questions. A recent World Bank research project (led by Demirgüç-Kunt), however, furnished both a database for researchers worldwideand the answers to several key questions on the impact of adopting ex-plicit deposit insurance on financial sector stability, the ability for mar-kets to exert discipline on banks, and the development of the overallfinancial system. In the process, conclusions on key design issues forauthorities are emerging.

The weight of evidence from this research is surprisingly clear cut, sug-gesting that in practice, rather than lowering the likelihood of a crisis, theadoption of explicit deposit insurance has been associated on average withless banking sector stability, and this result does not appear to be driven byreverse causation. Here the qualification “on average” is key: deposit insur-ance shows no significant destabilizing effect in countries with strong

—but may cause economicdamage—

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04110

111


institutions; only where the institutional environment is weak do prob-lems arise. The natural interpretation of this result is that banks, exploit-ing the availability of insured deposits, take greater risks. The presence ofexplicit insurance reduces depositor monitoring, and this matters if offi-cial supervision is insufficient, as where institutions are weak. The role ofgood institutions—as measured in this research by indicators of the rule oflaw, good governance (a proxy for effective regulation and supervision),and low corruption—thus seems crucial in reducing the opportunities forrisk-taking (Demirgüç-Kunt and Detragiache 2000).

That explicit deposit insurance could be positively correlated withbanking crises should not be considered too surprising, because when itis credible, it facilitates deposit gathering by banks regardless of the risksthey undertake.20

Even without explicit insurance, depositors could infer an implicitgovernment protection. At lower levels of institutional development,however, confidence in such implicit insurance may be low. There is nocertainty at all that the government will, in the event of a failure, be ableor willing to pay out even to small depositors, let alone large depositorsand shareholders. This uncertainty keeps depositors motivated to moni-tor banks (to the extent that they can), especially given that they cannotrely on strong official supervision of the banks in an environment ofpoor skills, a weak information and regulatory base, and often politicalinterference. In contrast, the announcement of an explicit scheme actslike a signal that bailout funds will be easier to get, even from a govern-ment operating in a weak institutional setting.21,22

Although these remarkable econometric findings have not, of course,gone unchallenged, it has so far proved impossible to dismiss them. True,in a recent working paper Eichengreen and Arteta (2000, pp. 44–45) con-tend that there is “at least as much evidence that depositinsurance…provides protection from depositor panics…as that it destabi-lizes banking systems.” In arriving at this conclusion, however, they con-fine themselves to a more limited sample of countries and crises. In par-ticular, by omitting countries with better institutions, they are unable todetect the importance of institutional quality in determining the overalleffectiveness of deposit insurance, as well as of different design features.When they widen their database, they confirm the above results.

Confirmation of the adverse impact of explicit deposit insurance onmarket discipline can be seen in the price that banks have to pay for theirdeposits. Examination of individual bank accounts shows that illiquid bankstend to pay more for their funds, partly reflecting depositors’ concern to

—by encouraging risk-taking in institutionallyweak settings

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04111


112

ensure their own liquidity, but the premium on interest expense for illiq-uid banks is less if a generous deposit protection system is in place. Inter-estingly, these findings come from a different cross-country database fromthat used in assessing the link with crises and, as such, provide importantadditional evidence. Inasmuch as they draw on individual balance sheetand income statement data from some 2,500 banks in up to 43 countries,this may be more telling direct evidence of the way in which deposit insur-ance can affect incentives (Demirgüç-Kunt and Huizinga 2000b). Althoughdeposit insurance weakens market discipline even in advanced countries,the effects seem to be offset by better official oversight and still more effec-tive market monitoring.

Martinez-Peria and Schmukler (2001) also found similar evidence inArgentina (in the early days after adopting explicit insurance), Chile,and Mexico of the market disciplining risky banks by demanding higherinterest rates. Interestingly, though, in this case even insured depositorsdisplayed some disciplining effect, which may represent a lack of cred-ibility toward the insurer’s commitment or speed in paying out.23 Still,where deposit insurance appeared most credible (in Chile), uninsureddepositors appeared to be more effective monitors of bank risk.

The lower interest rates point to the advantages gained by bank share-holders from the existence of deposit insurance, a gain that, in aggre-gate, is rarely paid for through insurance premia. “Correct” pricing wouldremove this subsidy, but it appears that it is easier to adopt deposit insur-ance than to price it correctly—and correct pricing is difficult in manyemerging markets. If the value of bank equities as quoted on an efficientstock market truly reflects the risks and returns facing the bank’s share-holders, it is possible to infer the ex ante value of the deposit insurancescheme to each bank by examining the leverage of the bank and thevariance of its stock price (box 2.5). The calculated values can be sub-stantial, and this tool could be used by supervisors to predict bank fail-ure, as Kaplan (1999) showed for Thailand.

Contrary to a popular view that deposit insurance might be needed inpoor countries to give the confidence to allow the financial deepeningthat is needed (cf. chapter 1) to support growth, the data suggest that, ininstitutionally weak environments, having explicit deposit insurance leadsto less financial sector development (Cull, Senbet, and Sorge 2000). Al-though it may be paradoxical that the provision of insurance could leadto less of an activity, it may be that when taxpayers in institutionally weakcountries see their authorities providing explicit guarantees, they under-stand that the environment is not conducive to restraining the cost of

Deposit insurance schemesmay inhibit financial sector

development whereinstitutions are weak

02 Chap02 Ed.1a (noRed) 3/19/01, 9:04112

113


these guarantees. The result, then, might be that the real insurers, thetaxpayers themselves, choose to hide their assets outside the banking sys-tem, and perhaps outside the country, to avoid being taxed for coverage.

When an explicit insurance system is adopted, the government takesover some of the monitoring function of banks. This requires both trans-parency—the ability to detect as well as possible the risks that bankersare taking—and deterrence—the ability to convince bankers that ruleswill be enforced. Deterrence in turn depends on the accountability ofgovernment officials, in particular those in the deposit insurance andrelated regulatory agencies (Kane 2000). Better levels of institutionaldevelopment—in the legal systems, accounting and auditing standards,and the political environment or quality of government—will make it

A BANK WHOSE DEPOSITS ARE INSURED CAN ACCESS

such deposits at close to the market price for risk-free deposits regardless of the risk it is taking on theasset side of its portfolio. Some of the risk, however,is passed through to shareholders, and in an efficientequity market, the price of a risky bank’s equity willbe lower on average and more volatile. Employingstandard arguments from the theory of option pricing,it is possible to infer from the volatility and level ofthe equity price, the market’s beliefs about theprobability of the bank failing and of the insurerhaving to pay out.

Using these probabilities, we can calculate the an-nual implicit subsidy—or expected insurance payout—for each bank. Although the formula is complex, onlythree variables are needed for this calculation, the eq-uity volatility, the ratio of equity to deposits, and thedividend yield. The following table presents a ready-reckoner allowing the implicit annual subsidy value tothe shareholders of deposit insurance for any bank givenonly the equity volatility and the ratio of equity todeposits. (The table assumes zero dividend yield.) Riskybanks—those with relatively little equity and volatileearnings—enjoy a large subsidy.

Box 2.5 Implicit value of deposit insurance to the bank’s shareholders

Annual implicit safety net subsidies as a percentage of the market value of equity

E/DσE

50 60 70 80 90 100

1 0.5 1.6 4.1 8.5 16.6 29.12 0.5 1.6 4.0 8.4 15.6 27.95 0.4 1.4 3.4 7.4 13.3 24.710 0.4 1.3 3.0 6.5 12.2 20.620 0.3 1.0 2.4 5.0 9.5 15.750 0.1 0.5 1.2 2.7 5.0 8.5

Note: σE is percentage annual volatility (standard deviation) of equity returns, E/D is the market value of the bank’s equity as apercentage of the value of the bank’s deposits. The dividend yield is assumed to be zero.

Source: Laeven (1999).

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05113


114

more difficult for bankers to gamble with insured deposits, or for gov-ernment officials to refrain from disciplining them.

So if we combine these three features—transparency, accountability,and deterrence—into the overall “institutional environment,” the argu-ment can be summarized in figure 2.9. Deposit insurance—whetherexplicit or implicit—provides the social benefit of protecting insureddepositors, but at the expense of socially costly moral hazard behavior.We can picture the level of depositor protection provided by a function-ing explicit system (the top panel) as being a given, independent of theremainder of the institutional environment. With an implicit one, somelevel of social protection usually will be provided, depending on whatthe government wants and is able to provide ex post. This may, however,as pictured in figure 2.9, be somewhat larger in countries that have

Figure 2.9 Deposit insurance: net benefits

Institutional environment

Social protection benefits

Implicit insurance

Explicit insurance


Moral hazard costs

Implicit insurance

Explicit insurance

Deposit insurance schemesbalance the social benefits of

security...

...against the social costs ofmoral hazard.

(figure continues on following page)

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05114

115


When institutions are weak, thecosts of explicit depositinsurance outweigh the benefits.

Source: See text.


Balance of advantage: explicit over implicit insurance

Figure 2.9 (continued)

achieved a higher overall institutional quality, if only because the better-developed tax systems there will permit greater coverage.

At low levels of institutional development, moral hazard behavior (themiddle panel) can run rampant with an explicit system—bankers willhave access to deposits, thanks to the insurance, but with weaker over-sight. This opportunistic behavior, however, will tend to be reduced witha better institutional environment. In contrast, when the environmentis weak, there likely is little moral hazard with an implicit system, asdepositors will expect little protection—indeed, they may keep theirwealth outside the banking system and even outside the country.

The bottom panel sums up the net benefits, with the key messagethat adequate infrastructure for enforcing contracts is of paramountimportance for ensuring net gains from explicit deposit insurance. Al-though it is not evident at what cutoff point explicit deposit insurancemight yield a net gain to a country, the need to do an “audit” of thestate of transparency, deterrence, and accountability prior to its adop-tion is clear. Governments at the low end of this spectrum that want toinstitute an explicit system should first focus on improving the relatedinstitutions—including the regulatory environment (discussed below)in order to reduce the likelihood of excessive risk-taking. Importantly,no evidence exists that there is any cost to waiting to adopt depositinsurance. In addition to the evidence noted here, that deposit insur-ance in weak environments tends to lower financial development (andthus growth), all high-income countries reached that stage without ex-plicit deposit insurance.

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05115


116

When authorities determine that their system is appropriate for ex-plicit deposit insurance, certain design features should be kept in mind.One way to determine design is to look just at industrial countries andfollow what they do, or otherwise try to infer best practice from firstprinciples (Garcia 1999). Moreover, the Financial Stability Forum’sWorking Group on Deposit Insurance has been asked to develop guid-ance on deposit insurance to assist countries that are adopting or signifi-cantly reforming a deposit insurance system, and a report is expected inthe fall of 2001. Wide differences, however, exist in the design of indus-trial country systems. More important, success may depend on replicat-ing other institutional features of advanced countries as well.

Another method to complement this approach would be to look atlessons derived from cross-country experience. The econometric findingsof Demirgüç-Kunt and Detragiache (2000) and Demirgüç-Kunt andHuizinga (2000b), already discussed above, and based on data from a widerange of countries, also point to several features of explicit schemes thatcan influence the degree to which they weaken market discipline or in-crease the risk of crisis, in particular, coverage, governance, and funding.

Coverage: The results suggest keeping coverage as low as is consistentwith the perceived need to protect small depositors.

There is room for disagreement on what the ceiling should be, but arule of thumb suggests a figure of one to two times annual per capita GDPas sufficiently generous to protect small depositors while still maintainingsignificant market discipline. Interbank deposits should be excluded.

Governance: Involving the private sector in the management and ad-ministration of the fund can help limit the reduction in market disci-pline and the impact on systemic risk, but is no cure-all.

This issue of governance has received less attention recently, but the keyrole of private involvement in mutual bank guarantees was at the heart ofsuccessful deposit protection systems in the early days. Mutual guaranteesare to be found, for example, in such successful mid-19th century U.S.state-based systems as in Indiana, Iowa, and Ohio (all of which featuredunlimited mutual liability and were relatively successful—White 1997),and in the clearinghouse associations in the 19th and early 20th century. Itis also a feature of several current deposit insurance systems, most notablythat of Germany. Private sector involvement and even responsibility fordeposit insurance illustrates the principle of the government harnessing theprivate sector to achieve its ends. Purely public schemes are more prone tocrisis, and they reduce market discipline, but private sector deposit schemesat times have failed, and they can run out of funds in a systemic crisis. 24

Don’t just copy a depositinsurance scheme from

another country

Limit coverage—

—involve the privatesector in sharing the

risk—

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05116

117


Importantly, it is easier to achieve private sector involvement in name,but without the exceptional oversight that characterizes some cases, suchas the German system. Thus, private systems appear to work best in thepresence of mutual liability and are best conceived of as a first round ofdefense against all but systemic crises, at which point the governmentcan step in—much as the risk against catastrophic loss against earth-quakes or hurricanes is handled.

The second potential drawback is that private schemes are based onpeer monitoring, which (as observed by Calomiris 1992) is more likely towork when the coalition is relatively limited in numbers. Beyond somepoint, members may be tempted to “free ride” on the monitoring of oth-ers. In the German system this problem is addressed by the existence ofseveral deposit insurance systems for different groups of banks. Smallernumbers of banks may also promote safety by boosting their franchisevalue and accordingly providing bankers with greater incentives to behaveprudently. Still, private coalitions could be used to stifle competition, andgovernments may have to make a decision on where to draw the line be-tween competition and stability. The high costs of banking crises in devel-oping countries suggests giving greater weight to stability. Moreover, manydeveloping countries, in particular the smaller ones, already have a rela-tively small number of banks, compared with those in their industrialcounterparts. Also, as is suggested in chapter 4, firms and households arerapidly gaining access to financial services from abroad, so that finance isbecoming more competitive even in small countries.

Finally, deposit protection systems like those in Germany may be suc-cessful because of the institutional and regulatory environment in whichthey reside. The strong antibankruptcy bias of German law and the effec-tive regulatory and supervisory system likely are important as well.25 Inter-estingly, applying the methodology of box 2.2 to a sample of 12 countries,Laeven (2000) concludes that German banks take the very low risks, andhave the lowest gross subsidy from deposit insurance. Private management,mutual liability, and the antibankruptcy bias likely explain this result.

Funding: The regression results introduce the possibility that keepingthe scheme unfunded, though with access to funds, may help protect marketdiscipline. Funding likely increases confidence that payout will be prompt.The case against funding, though, is controversial and not conclusive. TheU.S. savings and loan crisis showed that unfunded (or underfunded)schemes could result in greater forbearance and higher-cost resolutions asthe insurer struggled to protect depositors of weak banks. In addition, it issometimes argued that the decision to fund deposit insurance may be

—and keep schemesunfunded, or with muchoversight, in a weakinstitutional environment

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05117


118

accompanied by better oversight. Nevertheless, the cross-country econo-metrics points to the fact that funds can be abused more easily in weakinstitutional environments, and it seems far easier to set up a fund than toprotect it from looting. These findings should be borne in mind by au-thorities considering whether or not to fund. Leaving the scheme unfunded,but with the ability to access funds from the government, should allow aquick response while permitting oversight to minimize abuse. Ex ante fund-ing should only be considered when legal and regulatory institutions aredeveloped sufficiently to prevent looting.

In sum, authorities considering the adoption of deposit insurance canbenefit from these lessons. Some may interpret the evidence to mean thatif countries adopt a “good” deposit insurance system, they will be betterinsulated against crisis. The difficulty, however, is that the adoption ofdeposit insurance per se is a “stroke-of-the-pen reform,” and the institu-tional building to ensure that the system is “good” takes considerable time.Without adequate institutional development, the risk that deposit insur-ance could lead to crises, less financial sector development, more poorlyfunctioning financial markets, and ultimately slower growth and higherpoverty levels is real. Thus, authorities considering deposit insurance shouldmake an audit of their institutional framework the first step in thedecisionmaking process. Countries that do decide to establish an explicitdeposit insurance system should draw on these results of experience, whichutilize known market forces to ensure prudence.

Conclusions

THE CONSISTENT MESSAGE OF THIS CHAPTER IS THUS THAT

bank owners and other market participants should be viewedas necessary complements to official supervisors in monitoring

banks. Whatever the prudential regulations that are put in place—andit may be that more is needed than simply focusing on capital adequacy(cf. Honohan and Stiglitz 2001)—ensuring compliance is the majorstumbling block. Given information problems and the difficulty inunderstanding well how incentives are functioning, excessive weighton one group as the principle monitor is akin to excessive concentrationin a bank’s portfolio. It may appear to pay off nicely until failingmiserably. The strategic approach for authorities is to use incentiveswherever they can be applied to maximize the number of motivated,watchful eyes.

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05118

119


Easy access to an implicit or explicit safety net confers a subsidy onbanks, which encourages excessively bank-dependent—and debt-intensive—economies. Putting in place the recommendations of this chap-ter and effectively eliminating or greatly reducing this subsidy will removethis distortion and encourage the nonbank financial sector to develop. Tobe sure, there are some risks here, to the extent that it is near-bank activityjust outside the scope of the regulations that occurs, and regulatory designneeds to adaptive to avoid such arbitrage opportunities. To the extent,however, that it allows the emergence of nonbank types of finance, includ-ing market-traded equity and bonds, and the associated collective savingsinstitutions and other financial services activities, this will help the alloca-tion of risks and lower the cost of risk capital. Risk and fraud are present innonbank finance, too, but the existence of risk is known to all participantsand is rewarded by higher expected returns. Fraud needs to be dealt withthrough responsible disclosure standards and stiff penalties, as well as someconsumer-oriented regulations. With a safer banking system in place, au-thorities will be better able to avoid going down the dangerous road ofextending the safety net beyond banking.

There is no doubt that concentration of ownership and control, noted inchapter 1, can limit the efficacy of nonbank financial institutions and mar-kets in providing independent sources of finance and independent checkson the powers of powerful interests. Along with increasing access to foreignfinancial services (chapter 4), however, broadening capital markets over timepromises to provide greater diversity and stability to the financial sector.Improvements to basic financial infrastructure—enhancing disclosure andimproving the protection of shareholders and creditors, as noted in chapter1—will be instrumental in this task. To be sure, these recommendationsmay be difficult to implement, because politicians will need all their skills incombating powerful interests. Developing an awareness in society of thecosts that many, including the poor, must pay for a weak incentive environ-ment should help bolster support for improvements in the framework. Theforces of globalization (chapter 4) may help in this effort.

Notes

1. Kahneman and Tversky’s (1979) Prospect Theoryholds that individuals’ assessments of gains and losses canvary depending on their initial situation and specificallymay be averse to losses or loss realization, such as notselling stock whose prices fall.

2. As Kindleberger (1996, p. 66) notes, “...the pro-pensities to swindle and be swindled run parallel to thepropensity to speculate during a boom...And the signalfor panic is often the revelation of some swindle, theft,embezzlement, or fraud.”

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05119


120

3. Bagehot (1873, p. 131) reminds that during theSouth Sea Bubble, one of the companies whose shares werequoted was a bit peculiar. “But the most strange of all, per-haps, was ‘For an Undertaking which shall in due time berevealed.’ Each subscriber was to pay down two guineas,and hereafter to receive a share of one hundred, with adisclosure of the object; and so tempting was the offer, that1,000 of these subscriptions were paid the same morning,with which the projector went off in the afternoon.”

4. As Levine (1997) notes, Hicks [[ARE YOU SUREYOU WANT TO DELETE THIS ONE? IT IS IN-CLUDED IN THE REF. LIST.]] concluded that al-though the products in the early stages of the industrialrevolution were invented several decades earlier, their large-scale manufacture had to await the financial revolutionthat permitted the financing of illiquid investments.

5. Bernanke (1983) documented the credit channelfor the Great Depression of the 1930s.The role of a sup-ply-driven “credit crunch” in exacerbating the East Asiacrisis has been extensively debated (a representative col-lection of the research literature is in Domac, Ferri, andKawai forthcoming). To the extent that a summary con-clusion can be drawn, it appears that, while an acute creditsqueeze affected firms, especially SMEs in the early stageof the crisis, the economic downturn soon meant thatdemand for credit also declined, and relaunching creditsupply no longer seemed to be the most pressing issue—though scholars will remain divided on the degree to whichit did remain a problem. For the future, the priority willbe to ensure that both macropolicy stability and the regu-latory environment will be sufficiently secure to make dis-cussion of forbearance and subsidies unnecessary.

6. If three outliers are discarded, the correlation is0.7 and a regression line implies an approximate one-to-one relationship between flow output costs and fiscal costs.This finding could be interpreted as suggesting that thedifferent elements of cost are all correlated, and as sup-porting the use of fiscal cost as a general-purpose approxi-mation to the unobserved total economic cost.

7. Other goals, such as antidiscrimination and pro-motion of home ownership and of exports continue to bepursued through detailed measures of financial policy insome countries, but these will not be discussed here. There

has been a decline in the perceived effectiveness of policymeasures that seek to direct the flow of finance to specificeconomic goals (Caprio, Hanson, and Honohan 2001).

8. Official action to help prevent the outright fail-ure of the highly leveraged hedge fund LTCM in 1998was substantially driven by knowledge of the potentialimpact of such a failure on the stability of the bankingsystem (reference).

9. The ending of liquidity requirements in develop-ing countries came about in emulation of the new, bestpractice in the OECD area, and lower liquidity require-ments did alleviate somewhat the taxation of the finan-cial sector. Although liquidity ratios—holdings of centralbank reserves, cash, and government paper—were notneeded for prudential purposes in high-income countries,developing countries have not been able to upgrade banksupervision and regulation sufficiently to offset the lossof this buffer, cf. Caprio and Honohan (2000).

10. Overly simple or inflexible rules can have unfor-tunate side effects. In a downturn, for instance, rigid bankcapital requirements can accentuate the recession by con-straining credit growth, especially if banks have to provi-sion more against loan losses (Chiuri, Ferri, and Majnoni2000). However, the theoretical solution of making thecapital requirements explicitly cycle-dependent(Dewatripont and Tirole 1993) may, in practice be hardto implement credibly or without risking a degree of for-bearance that could altogether undermine the incentiveeffect of having capital requirements.

11. It is not only emerging economies thatunderprovision. A recent Bank of Japan study (1998)found that 75.3 percent of loans classified in 1993–94 asdoubtful at a sample of 18 banks became write-offs overthe following three years—but required provisioning forsuch loans is only 52 percent; and that 16.7 percent of“category 2” loans, for which only a 2 percent provisionis required, were written off.

12. The integration of financial sector supervision hasreceived much attention, but is beyond the scope of thisstudy. As integrated agencies are relatively recent, no for-mal quantitative research of their relative merits has beenperformed, and only anecdotal information (such as the

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05120

121


continued difficulty in getting effective cooperation be-tween separate departments in a single agency) is avail-able. Still, as Goodhart (2000) argues, for emerging mar-kets this issue is premature and likely of second order rela-tive to fixing the overall incentive environment.

13. As Becker and Stigler (1974) note, “The appropri-ate pay structure has three components: an ‘entrance fee,’equal to the temptation of malfeasance, a salary premiumin each year of employment approximately equal to theincome yielded by the ‘entrance fee,’ and a pension with acapital value approximately equal also to the temptation ofmalfeasance. As it were, enforcers post a bond equal to thetemptation of malfeasance, receive the income on the bondas long as they are employed, and have the bond returnedif they behave themselves until retirement. Put differently,they forfeit their bond if they are fired for malfeasance.”

14. On the other hand, it is recently reported thatsenior executives of Daiwa Bank have been held person-ally liable for losses caused by an inadequately supervisedtrader (Economist, October 6, 2000 [[OR NOVEMBER16, AS IN REF. LIST?]]).

15. In some transitional cases, authorities may havebeen partly motivated by the possibility of EuropeanUnion (EU) accession and the agreed model for depositinsurance there.

16. Of course, since “big money” also is “smartmoney,” it may run first, and to the extent that authori-ties are concerned about a potential “systemic” crisis, theymay elect to cover uninsured and large depositors, evenincluding interbank claims, either through the depositinsurance fund or some other facility. Thus during theContinental Illinois difficulties in the United States, de-posit insurance was extended to all creditors.

17. At times, governments have exceeded their owncoverage limits, but the empirical findings recounted be-low show that having lower ceilings does seem to matter.

18. The requirement in EU law for member states tocover a common euro amount of deposits has placed up-

ward pressure on coverage levels in countries aspiring toEU membership.

19. In the United States, as in some countries, thereis a limit on the total amount of funds in the depositinsurance fund. Once that limit is reached, banks are nolonger assessed until funds drop below the ceiling. In thissituation, banks face a zero premium, and clearly no riskdifferentiation.

20. Demirguc-Kunt and Detragiache (1999) found—although in a small sample of 24 countries—that the costof crises also was higher with deposit insurance and weakinstitutional environments.

21. Similar arguments have been made in regard toforeign exchange reserves.

22. Any message that the coverage will be limitedseems to be discounted in such institutional settings.

23. The latter can be significant in emerging mar-kets, where it has taken from months to as long as eightyears for depositors to be paid in accordance with depositinsurance statutes.

24. Neither private nor public systems, however, weredesigned for systemic crises, but rather to prevent epi-sodes of individual bank failure from mushrooming intoa systemic problem.

25. According to the La Porta and others (1997) da-tabase, Germany ranks among the highest in the protec-tion of creditors’ rights. Also, as Beck (2000) reports, al-though only fraudulent bankruptcy is subject to prosecu-tion, in Germany fraud can include violating “orderly busi-ness practice,” which can be broadly interpreted. HansGerling, a principal of Herstatt Bank, contributed about150 million DM to creditors to avoid legal entanglementsafter that bank failed. Moreover, the German BankingAct prohibits any manager involved in fraudulent bank-ruptcy from ever holding a managerial post in banking—as determined by regulatory officials rather than criminalprosecution.

02 Chap02 Ed.1a (noRed) 3/19/01, 9:05121

Date post:	11-Mar-2020
Category:	Documents
Upload:	others
View:	2 times
Download:	0 times

Preventing and Minimizing...

Documents